Fuel card fraud is caught at the transaction level, by comparing every purchase against what the truck was actually doing at that moment. Card controls alone will not find it, because a compromised card used within its limits at an approved merchant looks exactly like a legitimate fill. What finds it is running each transaction against the trip, the tank, the clock and the driver's history: gallons exceeding tank capacity, two fills too far apart for the time between them, the same card twice at one station minutes apart, purchases outside the driver's duty window, fills with no active load assigned, and purchases well off the planned route. ValveRide Flow runs eight such detectors on every transaction a fleet posts. The output is a review queue, not an accusation. Most hits have innocent explanations, which is the point: you want the short list of things worth a look, not a verdict.
Updated August 2026 · 8-minute read
Every one of these runs on every transaction a fleet posts. Each surfaces transactions to review rather than reaching a conclusion.
Fuel card fraud is caught at the transaction level, by comparing every purchase against what the truck was actually doing at that moment. Card controls alone will not find it, because a compromised card used within its limits at an approved merchant looks exactly like a legitimate fill. What finds it is running each transaction against the trip, the tank, the clock and the driver's history: gallons exceeding tank capacity, two fills too far apart for the time between them, the same card twice at one station minutes apart, purchases outside the driver's duty window, fills with no active load assigned, and purchases well off the planned route. ValveRide Flow runs eight such detectors on every transaction a fleet posts. The output is a review queue, not an accusation. Most hits have innocent explanations, which is the point: you want the short list of things worth a look, not a verdict.
Common enough that most fleets above a handful of trucks find something when they first look, but the something is usually not dramatic. Compromised card numbers, friends-and-family fill-ups, and personal-vehicle fills are the recurring patterns. Just as often the first real finding is not theft at all, it is a discount that stopped applying months ago, or a tank capacity recorded wrong on a truck. Both cost money and neither shows up on a spend report.
Some, deliberately. A detector tuned to fire only when it is certain will miss the cumulative patterns that matter most, and those patterns are the ones that add up. The design goal is a short ranked review queue rather than either an alarm or a verdict, with the reason for each flag visible so a manager can dismiss an obvious explanation quickly. Thresholds like off-route also require sustained signal rather than a single ping, which removes most of the noise.
No, and you generally should not start there. Most flags resolve into a data problem or a routine explanation. The productive posture is to treat the queue as a list of things worth understanding: a wrong tank capacity gets corrected, a broken discount feed gets fixed with the vendor, a repeat-station pattern becomes a coaching conversation, and only a small residue looks like anything else. Fleets that open with accusations tend to stop using the tool.
The fuel card transactions and enough operational context to judge them: the trips, the truck and tank details, and hours-of-service or position data where available. In practice this is the same data Flow already uses to build fuel plans, which is why detection comes with optimization rather than as a separate product. A fleet running Flow for planning has the inputs by definition.
Anomaly Watch is part of Growth and Enterprise rather than a standalone product, because it depends on the same route, tank and contract data the optimizer builds. If detection is your only interest, that is worth saying in a demo so the conversation stays on it.
Related: getting discounts onto every load, how to audit a savings claim, how Flow works.
A 30-minute demo can walk your recent fuel card data through Anomaly Watch and show what the queue looks like for your fleet, including the boring findings that turn out to be worth money.